Datasets:
id stringlengths 10 10 | image imagewidth (px) 1.02k 2.45k | question stringclasses 391 values | answer stringlengths 1 18 | type stringclasses 24 values | tier int32 1 3 | gsd float32 0.3 1 | acceptable_range_lower float64 0 146 ⌀ | acceptable_range_upper float64 0 153 ⌀ |
|---|---|---|---|---|---|---|---|---|
SQuID_0000 | What is the total vegetation area (in hectares) within 200m of barren land? (GSD: 0.5m) | 105.55 | proximity_area | 2 | 0.5 | 103.18 | 107.92 | |
SQuID_0001 | What percentage of the image is urban area within 500m of vegetation? (GSD: 0.5m) | 34.35 | proximity_percentage | 2 | 0.5 | 32.1 | 36.6 | |
SQuID_0002 | Find vegetation patches larger than 2 hectares, then calculate how much of their area (in hectares) falls within 200m of water bodies (GSD: 0.5m) | 63.93 | complex_vegetation_water_access | 3 | 0.5 | 62.49 | 65.37 | |
SQuID_0003 | How many separate agricultural land patches between 0.125 and 10 hectares are there? (GSD: 0.3m) | 2 | connectivity | 2 | 0.3 | 1 | 3 | |
SQuID_0004 | Is there more barren land than forest area in this image? (GSD: 0.3m) | 1 | binary_comparison | 1 | 0.3 | null | null | |
SQuID_0005 | What percentage of the image is covered by the largest vegetation region (among regions larger than 0.125 hectares)? (GSD: 0.5m) | 75.81 | size | 1 | 0.5 | 74.08 | 77.55 | |
SQuID_0006 | How many separate urban area regions are there? When counting, ignore patches smaller than 0.1 hectares. (GSD: 0.5m) | 4 | count | 1 | 0.5 | 3 | 5 | |
SQuID_0007 | How many buildings (larger than 0.01 hectares) are within 500m of agricultural land? (GSD: 0.3m) | 4 | building_proximity | 2 | 0.3 | 3 | 5 | |
SQuID_0008 | Find barren land patches larger than 1 hectares, then calculate how much of their area (in hectares) falls within 200m of forest area (GSD: 0.3m) | 1.49 | complex_multi_condition | 3 | 0.3 | 1.46 | 1.52 | |
SQuID_0009 | Find urban patches larger than 1 hectare, then calculate how much of their area (in hectares) falls within 100m of water bodies (flood risk assessment) (GSD: 0.5m) | 4.03 | complex_urban_flood_risk | 3 | 0.5 | 3.94 | 4.12 | |
SQuID_0010 | What percentage of the image is covered by barren land? (GSD: 0.3m) | 23.36 | percentage | 1 | 0.3 | 21.62 | 25.09 | |
SQuID_0011 | How many separate agricultural land regions are there? When counting, ignore patches smaller than 0.125 hectares. (GSD: 0.3m) | 4 | count | 1 | 0.3 | 3 | 5 | |
SQuID_0012 | What percentage of the image is forest area within 50m of agricultural land? (GSD: 0.3m) | 9.55 | proximity_percentage | 2 | 0.3 | 7.3 | 11.8 | |
SQuID_0013 | What percentage of the image is covered by forest area? (GSD: 0.3m) | 3.09 | percentage | 1 | 0.3 | 1.35 | 4.83 | |
SQuID_0014 | Find water bodies patches larger than 5 hectares, then calculate how much of their area (in hectares) falls within 200m of barren land (GSD: 0.3m) | 5.54 | complex_multi_condition | 3 | 0.3 | 5.42 | 5.66 | |
SQuID_0015 | What percentage of the image is covered by the largest water bodies region (among regions larger than 0.1 hectares)? (GSD: 0.3m) | 13.7 | size | 1 | 0.3 | 11.96 | 15.43 | |
SQuID_0016 | What percentage of the image is water bodies within 100m of agricultural land? (GSD: 0.3m) | 8.33 | proximity_percentage | 2 | 0.3 | 6.08 | 10.58 | |
SQuID_0017 | Find urban area patches larger than 1 hectares, then calculate how much of their area (in hectares) falls within 500m of water bodies (GSD: 0.5m) | 3.02 | complex_multi_condition | 3 | 0.5 | 2.95 | 3.09 | |
SQuID_0018 | Is there more vegetation than urban area in this image? (GSD: 0.5m) | 1 | binary_comparison | 1 | 0.5 | null | null | |
SQuID_0019 | Find agricultural land patches larger than 1 hectares, then calculate how much of their area (in hectares) falls within 500m of forest area (GSD: 0.3m) | 4.95 | complex_multi_condition | 3 | 0.3 | 4.84 | 5.06 | |
SQuID_0020 | What percentage of the image is barren land within 50m of vegetation? (GSD: 0.5m) | 0.4 | proximity_percentage | 2 | 0.5 | 0 | 2.65 | |
SQuID_0021 | What is the area of the largest solar installation in hectares? (GSD: 0.3m) | 0.91 | size | 1 | 0.3 | 0 | 2.65 | |
SQuID_0022 | Is there any barren land within 100m of urban area? (GSD: 0.5m) | 1 | binary_proximity | 2 | 0.5 | null | null | |
SQuID_0023 | Find vegetation patches larger than 2 hectares, then calculate how much of their area (in hectares) falls within 200m of water bodies (GSD: 0.5m) | 22.91 | complex_vegetation_water_access | 3 | 0.5 | 22.39 | 23.43 | |
SQuID_0024 | Find vegetation patches larger than 5 hectares, then calculate how much of their area (in hectares) falls within 500m of barren land (GSD: 0.5m) | 78.1 | complex_multi_condition | 3 | 0.5 | 76.34 | 79.86 | |
SQuID_0025 | Find water bodies patches larger than 1 hectares, then calculate how much of their area (in hectares) falls within 200m of agricultural land (GSD: 0.3m) | 0 | complex_multi_condition | 3 | 0.3 | 0 | 0 | |
SQuID_0026 | Find agricultural land patches larger than 2 hectares, then calculate how much of their area (in hectares) falls within 200m of water bodies (GSD: 0.3m) | 4.47 | complex_agriculture_water_access | 3 | 0.3 | 4.37 | 4.57 | |
SQuID_0027 | Find urban area patches larger than 1 hectares, then calculate how much of their area (in hectares) falls within 200m of vegetation (GSD: 0.5m) | 46.65 | complex_multi_condition | 3 | 0.5 | 45.6 | 47.7 | |
SQuID_0028 | Find agricultural land patches larger than 1 hectares, then calculate how much of their area (in hectares) falls within 200m of barren land (GSD: 0.3m) | 3.67 | complex_multi_condition | 3 | 0.3 | 3.59 | 3.75 | |
SQuID_0029 | Find agricultural land patches larger than 1 hectares, then calculate how much of their area (in hectares) falls within 500m of water bodies (GSD: 0.3m) | 8.17 | complex_multi_condition | 3 | 0.3 | 7.99 | 8.35 | |
SQuID_0030 | Find vegetation patches larger than 2 hectares, then calculate how much of their area (in hectares) falls within 200m of water bodies (GSD: 0.5m) | 17.09 | complex_vegetation_water_access | 3 | 0.5 | 16.71 | 17.47 | |
SQuID_0031 | How many separate water bodies patches between 0.1 and 3 hectares are there? (GSD: 0.3m) | 2 | connectivity | 2 | 0.3 | 1 | 3 | |
SQuID_0032 | Is there more vegetation than water bodies in this image? (GSD: 0.5m) | 1 | binary_comparison | 1 | 0.5 | null | null | |
SQuID_0033 | How many buildings (larger than 0.01 hectares) are within 200m of water bodies? (GSD: 0.3m) | 16 | building_proximity | 2 | 0.3 | 12 | 20 | |
SQuID_0034 | Is there more barren land than forest area in this image? (GSD: 0.3m) | 1 | binary_comparison | 1 | 0.3 | null | null | |
SQuID_0035 | Is there any barren land within 500m of urban area? (GSD: 0.5m) | 1 | binary_proximity | 2 | 0.5 | null | null | |
SQuID_0036 | What percentage of the image is urban area within 500m of water bodies? (GSD: 0.5m) | 0.03 | proximity_percentage | 2 | 0.5 | 0 | 2.28 | |
SQuID_0037 | What is the total vegetation area (in hectares) within 200m of urban area? (GSD: 0.5m) | 19.03 | proximity_area | 2 | 0.5 | 18.6 | 19.46 | |
SQuID_0038 | How many separate agricultural land patches between 0.125 and 5 hectares are there? (GSD: 0.3m) | 2 | connectivity | 2 | 0.3 | 1 | 3 | |
SQuID_0039 | What is the total water bodies area (in hectares) within 50m of agricultural land? (GSD: 0.3m) | 0.11 | proximity_area | 2 | 0.3 | 0.11 | 0.11 | |
SQuID_0040 | How many buildings (larger than 0.01 hectares) are within 200m of water bodies? (GSD: 0.3m) | 8 | building_proximity | 2 | 0.3 | 6 | 10 | |
SQuID_0041 | What percentage of the image is covered by the largest agricultural land region (among regions larger than 0.125 hectares)? (GSD: 0.3m) | 40.66 | size | 1 | 0.3 | 38.92 | 42.39 | |
SQuID_0042 | Is the forest area connected or fragmented (more than 5 separate patches larger than 0.125 hectares)? (GSD: 0.3m) | connected | fragmentation | 2 | 0.3 | null | null | |
SQuID_0043 | Find agricultural land patches larger than 1 hectares, then calculate how much of their area (in hectares) falls within 500m of forest area (GSD: 0.3m) | 1.39 | complex_multi_condition | 3 | 0.3 | 1.36 | 1.42 | |
SQuID_0044 | Find agricultural land patches larger than 2 hectares, then calculate how much of their area (in hectares) falls within 200m of water bodies (GSD: 0.3m) | 6.92 | complex_agriculture_water_access | 3 | 0.3 | 6.76 | 7.08 | |
SQuID_0045 | How many separate urban area regions are there? When counting, ignore patches smaller than 0.1 hectares. (GSD: 0.5m) | 3 | count | 1 | 0.5 | 2 | 4 | |
SQuID_0046 | How many separate water bodies patches between 0.1 and 5 hectares are there? (GSD: 0.3m) | 1 | connectivity | 2 | 0.3 | 0 | 2 | |
SQuID_0047 | Is the water bodies connected or fragmented (more than 5 separate patches larger than 0.1 hectares)? (GSD: 0.3m) | connected | fragmentation | 2 | 0.3 | null | null | |
SQuID_0048 | Find agricultural land patches larger than 1 hectares, then calculate how much of their area (in hectares) falls within 500m of water bodies (GSD: 0.3m) | 3.54 | complex_multi_condition | 3 | 0.3 | 3.46 | 3.62 | |
SQuID_0049 | How many buildings are there in the image? When counting, ignore buildings smaller than 0.01 hectares. (GSD: 0.3m) | 4 | count | 1 | 0.3 | 3 | 5 | |
SQuID_0050 | What percentage of the image is agricultural land within 500m of water bodies? (GSD: 0.3m) | 28.04 | proximity_percentage | 2 | 0.3 | 25.79 | 30.29 | |
SQuID_0051 | Find agricultural land patches larger than 1 hectares, then calculate how much of their area (in hectares) falls within 500m of water bodies (GSD: 0.3m) | 8.02 | complex_multi_condition | 3 | 0.3 | 7.84 | 8.2 | |
SQuID_0052 | Is there more barren land than agricultural land in this image? (GSD: 0.3m) | 1 | binary_comparison | 1 | 0.3 | null | null | |
SQuID_0053 | Find vegetation patches larger than 1 hectares, then calculate how much of their area (in hectares) falls within 500m of urban area (GSD: 0.5m) | 90.91 | complex_multi_condition | 3 | 0.5 | 88.86 | 92.96 | |
SQuID_0054 | Is there more urban area than water bodies in this image? (GSD: 0.5m) | 1 | binary_comparison | 1 | 0.5 | null | null | |
SQuID_0055 | Is the forest area connected or fragmented (more than 5 separate patches larger than 0.125 hectares)? (GSD: 0.3m) | connected | fragmentation | 2 | 0.3 | null | null | |
SQuID_0056 | How many separate urban area regions are there? When counting, ignore patches smaller than 0.1 hectares. (GSD: 0.5m) | 13 | count | 1 | 0.5 | 10 | 16 | |
SQuID_0057 | How many separate water bodies patches between 0.1 and 5 hectares are there? (GSD: 0.3m) | 2 | connectivity | 2 | 0.3 | 1 | 3 | |
SQuID_0058 | Is the water bodies connected or fragmented (more than 5 separate patches larger than 0.1 hectares)? (GSD: 0.3m) | connected | fragmentation | 2 | 0.3 | null | null | |
SQuID_0059 | Is there any barren land within 50m of urban area? (GSD: 0.5m) | 0 | binary_proximity | 2 | 0.5 | null | null | |
SQuID_0060 | What percentage of the image is covered by the largest vegetation region (among regions larger than 0.125 hectares)? (GSD: 0.5m) | 99.79 | size | 1 | 0.5 | 98.06 | 100 | |
SQuID_0061 | What percentage of the image is agricultural land within 50m of water bodies? (GSD: 0.3m) | 17.43 | proximity_percentage | 2 | 0.3 | 15.18 | 19.68 | |
SQuID_0062 | What percentage of the image is covered by the largest agricultural land region (among regions larger than 0.125 hectares)? (GSD: 0.3m) | 15.22 | size | 1 | 0.3 | 13.48 | 16.96 | |
SQuID_0063 | Is there more urban area than barren land in this image? (GSD: 0.5m) | 1 | binary_comparison | 1 | 0.5 | null | null | |
SQuID_0064 | Find urban patches larger than 1 hectare, then calculate how much of their area (in hectares) falls within 50m of vegetation (fire risk assessment) (GSD: 0.5m) | 3.02 | complex_urban_fire_risk | 3 | 0.5 | 2.95 | 3.09 | |
SQuID_0065 | What percentage of the image is covered by the largest water bodies region (among regions larger than 0.1 hectares)? (GSD: 0.3m) | 37.69 | size | 1 | 0.3 | 35.95 | 39.42 | |
SQuID_0066 | Find agricultural land patches larger than 5 hectares, then calculate how much of their area (in hectares) falls within 200m of barren land (GSD: 0.3m) | 0 | complex_multi_condition | 3 | 0.3 | 0 | 0 | |
SQuID_0067 | Find agricultural land patches larger than 2 hectares, then calculate how much of their area (in hectares) falls within 200m of water bodies (GSD: 0.3m) | 4.39 | complex_agriculture_water_access | 3 | 0.3 | 4.29 | 4.49 | |
SQuID_0068 | Find vegetation patches larger than 5 hectares, then calculate how much of their area (in hectares) falls within 500m of barren land (GSD: 0.5m) | 128.09 | complex_multi_condition | 3 | 0.5 | 125.21 | 130.97 | |
SQuID_0069 | What percentage of the image is barren land within 50m of forest area? (GSD: 0.3m) | 8.52 | proximity_percentage | 2 | 0.3 | 6.27 | 10.77 | |
SQuID_0070 | What percentage of the image is covered by the largest water bodies region (among regions larger than 0.1 hectares)? (GSD: 0.3m) | 6.4 | size | 1 | 0.3 | 4.67 | 8.14 | |
SQuID_0071 | Find vegetation patches larger than 1 hectares, then calculate how much of their area (in hectares) falls within 500m of water bodies (GSD: 0.5m) | 142.1 | complex_multi_condition | 3 | 0.5 | 138.9 | 145.3 | |
SQuID_0072 | Are there any solar panels larger than 0.01 hectares in this image? (GSD: 0.3m) | 1 | binary_presence | 1 | 0.3 | null | null | |
SQuID_0073 | Find water bodies patches larger than 5 hectares, then calculate how much of their area (in hectares) falls within 500m of agricultural land (GSD: 0.3m) | 8.94 | complex_multi_condition | 3 | 0.3 | 8.74 | 9.14 | |
SQuID_0074 | Find vegetation patches larger than 1 hectares, then calculate how much of their area (in hectares) falls within 200m of urban area (GSD: 0.5m) | 29.85 | complex_multi_condition | 3 | 0.5 | 29.18 | 30.52 | |
SQuID_0075 | How many separate barren land patches between 0.125 and 5 hectares are there? (GSD: 0.3m) | 3 | connectivity | 2 | 0.3 | 2 | 4 | |
SQuID_0076 | What is the total vegetation area (in hectares) within 100m of water bodies? (GSD: 0.5m) | 16.42 | proximity_area | 2 | 0.5 | 16.05 | 16.79 | |
SQuID_0077 | What percentage of the image is water bodies within 500m of agricultural land? (GSD: 0.3m) | 5.18 | proximity_percentage | 2 | 0.3 | 2.93 | 7.43 | |
SQuID_0078 | Is there more water bodies than forest area in this image? (GSD: 0.3m) | 1 | binary_comparison | 1 | 0.3 | null | null | |
SQuID_0079 | What percentage of the image is covered by agricultural land? (GSD: 0.3m) | 20.95 | percentage | 1 | 0.3 | 19.21 | 22.68 | |
SQuID_0080 | Find agricultural land patches larger than 5 hectares, then calculate how much of their area (in hectares) falls within 200m of barren land (GSD: 0.3m) | 5.64 | complex_multi_condition | 3 | 0.3 | 5.51 | 5.77 | |
SQuID_0081 | Is there more barren land than forest area in this image? (GSD: 0.3m) | 1 | binary_comparison | 1 | 0.3 | null | null | |
SQuID_0082 | Is there any water bodies within 500m of barren land? (GSD: 0.3m) | 1 | binary_proximity | 2 | 0.3 | null | null | |
SQuID_0083 | What is the total forest area area (in hectares) within 50m of agricultural land? (GSD: 0.3m) | 0.16 | proximity_area | 2 | 0.3 | 0.16 | 0.16 | |
SQuID_0084 | Calculate the solar potential MW output assuming 200W/m² efficiency. (GSD: 0.3m) | 0.74 | power_calculation | 2 | 0.3 | 0.73 | 0.75 | |
SQuID_0085 | Is there more water bodies than forest area in this image? (GSD: 0.3m) | 1 | binary_comparison | 1 | 0.3 | null | null | |
SQuID_0086 | What percentage of the image is covered by the smallest barren land region (excluding patches smaller than 0.125 hectares)? (GSD: 0.3m) | 9.57 | size | 1 | 0.3 | 7.83 | 11.3 | |
SQuID_0087 | How many separate agricultural land regions are there? When counting, ignore patches smaller than 0.125 hectares. (GSD: 0.3m) | 2 | count | 1 | 0.3 | 1 | 3 | |
SQuID_0088 | What percentage of the image is covered by the largest agricultural land region (among regions larger than 0.125 hectares)? (GSD: 0.3m) | 52.84 | size | 1 | 0.3 | 51.11 | 54.58 | |
SQuID_0089 | What is the total water bodies area (in hectares) within 100m of forest area? (GSD: 0.3m) | 0.14 | proximity_area | 2 | 0.3 | 0.14 | 0.14 | |
SQuID_0090 | Find vegetation patches larger than 1 hectares, then calculate how much of their area (in hectares) falls within 200m of barren land (GSD: 0.5m) | 16.72 | complex_multi_condition | 3 | 0.5 | 16.34 | 17.1 | |
SQuID_0091 | Find agricultural land patches larger than 1 hectares, then calculate how much of their area (in hectares) falls within 200m of water bodies (GSD: 0.3m) | 6.17 | complex_multi_condition | 3 | 0.3 | 6.03 | 6.31 | |
SQuID_0092 | What percentage of the image is covered by the smallest urban area region (excluding patches smaller than 0.1 hectares)? (GSD: 0.5m) | 7.07 | size | 1 | 0.5 | 5.33 | 8.8 | |
SQuID_0093 | What percentage of the image is covered by the smallest barren land region (excluding patches smaller than 0.125 hectares)? (GSD: 0.3m) | 2.81 | size | 1 | 0.3 | 1.07 | 4.54 | |
SQuID_0094 | What percentage of the image is solar panels? (GSD: 0.3m) | 40.12 | percentage | 1 | 0.3 | 38.38 | 41.85 | |
SQuID_0095 | Find agricultural land patches larger than 1 hectares, then calculate how much of their area (in hectares) falls within 500m of forest area (GSD: 0.3m) | 8.34 | complex_multi_condition | 3 | 0.3 | 8.15 | 8.53 | |
SQuID_0096 | What is the total water bodies area (in hectares) within 100m of forest area? (GSD: 0.3m) | 0.56 | proximity_area | 2 | 0.3 | 0.55 | 0.57 | |
SQuID_0097 | Is there any water bodies within 200m of barren land? (GSD: 0.3m) | 1 | binary_proximity | 2 | 0.3 | null | null | |
SQuID_0098 | What percentage of the image is covered by the largest forest area region (among regions larger than 0.125 hectares)? (GSD: 0.3m) | 3.32 | size | 1 | 0.3 | 1.58 | 5.05 | |
SQuID_0099 | Is there more vegetation than urban area in this image? (GSD: 0.5m) | 1 | binary_comparison | 1 | 0.5 | null | null |
SQuID: Satellite Quantitative Intelligence Dataset
A comprehensive benchmark for evaluating quantitative spatial reasoning in Vision-Language Models using satellite imagery.
Related Resources
- Code repository: https://github.com/PeterAMassih/qvlm-squid
- Paper (arXiv): https://arxiv.org/abs/2601.13401
Dataset Overview
- 2000 questions testing spatial reasoning on satellite imagery
- 587 unique images across four datasets
- 1950 auto-labeled questions from segmentation masks (DeepGlobe, EarthVQA, Solar Panels)
- 50 human-annotated questions from NAIP imagery with consensus answers
- 1577 questions include human-agreement ranges for numeric answers
- 3 difficulty tiers: Basic (710), Spatial (616), Complex (674)
- 3 resolution levels: 0.3m, 0.5m, 1.0m GSD
Human Annotation & Agreement Methodology
Human Annotation Process
- 50 questions on NAIP 1.0m GSD imagery were annotated by humans
- 10 annotators per question resulting in 500 total annotations
- Answer aggregation:
- Numeric questions: Used MEDIAN of all responses for robustness
- Categorical questions (connected/fragmented): Used MAJORITY voting
- Binary questions: Converted yes/no to 1/0 and used majority
Human Agreement Quantification
From the 500 human annotations, we computed the Mean Median Absolute Deviation (MAD) for each question type:
- Percentage questions: MAD = ±1.74 percentage points
- Proximity questions: MAD = ±2.25 percentage points
- Count questions: Normalized MADc = 0.19 (proportional to count magnitude)
For count questions, we use a normalized MAD (MADc) that makes the acceptable range proportional to the count value:
MADc = median(|Xi - median(X)|) / median(X) = 0.19
Acceptable Range Calculation
These MAD values were applied to ALL numeric questions in the benchmark to define acceptable ranges:
import math
# For percentage questions (absolute deviation)
if question_type == 'percentage':
lower = max(0.0, answer - 1.74)
upper = min(100.0, answer + 1.74)
# For count questions (proportional deviation)
# range(C) = [C - max(1, C × MADc), C + max(1, C × MADc)]
elif question_type in ['count', 'building_proximity', 'building_flood_risk',
'building_fire_risk', 'connectivity']:
MADc = 0.19
dr = max(1, answer * MADc) # At least ±1 deviation
lower = max(0, math.floor(answer - dr))
upper = math.ceil(answer + dr)
# For proximity percentage questions (absolute deviation)
elif 'within' in question and 'm of' in question:
lower = max(0.0, answer - 2.25)
upper = min(100.0, answer + 2.25)
Example count ranges with MADc = 0.19:
- C=5 → range [4, 7]
- C=10 → range [8, 12]
- C=50 → range [40, 60]
- C=100 → range [81, 120]
Special cases:
- Zero values have no range (exact match required)
- Binary/fragmentation questions have no range (exact match)
- Ranges are capped at valid bounds (0-100 for percentages, ≥0 for counts)
Question Types
The benchmark includes 24 distinct question types organized into three tiers:
Tier 1: Basic Questions (710 questions)
- percentage: Coverage percentage of a land use class
- count: Number of separate regions or objects
- size: Area measurements of regions
- total_area: Total area covered by a class
- binary_comparison: Comparing quantities between two classes
- binary_presence: Checking if a class exists
- binary_threshold: Testing if values exceed thresholds
- binary_multiple: Checking for multiple instances
Tier 2: Spatial Questions (616 questions)
- proximity_percentage: Percentage of one class near another
- proximity_area: Area of one class near another
- binary_proximity: Presence of one class near another
- building_proximity: Number of buildings near other features
- building_flood_risk: Buildings at flood risk (near water)
- building_fire_risk: Buildings at fire risk (near forest)
- connectivity: Counting isolated patches by size
- fragmentation: Assessing if regions are connected or fragmented
- power_calculation: Calculating solar panel power output
Tier 3: Complex Questions (674 questions)
- complex_multi_condition: Areas meeting multiple spatial criteria
- complex_urban_flood_risk: Urban areas at flood risk (near water)
- complex_urban_fire_risk: Urban areas at fire risk (near forest)
- complex_agriculture_water_access: Agricultural land with irrigation potential
- complex_size_filter: Filtering by size thresholds
- complex_average: Average sizes of regions
Loading the Dataset
from datasets import load_dataset
# Load dataset
dataset = load_dataset("PeterAM4/SQuID")
# Access a sample
sample = dataset['train'][0]
image = sample['image'] # PIL Image
question = sample['question']
answer = sample['answer'] # String or numeric
type = sample['type']
# Convert answer based on type
if type in ['percentage', 'count', 'proximity_percentage', 'proximity_area',
'building_proximity', 'building_flood_risk', 'building_fire_risk',
'connectivity', 'size', 'total_area', 'power_calculation'] or 'complex' in type:
answer_value = float(answer)
elif 'binary' in type:
answer_value = int(answer) # 0 or 1
elif type == 'fragmentation':
answer_value = answer # "connected" or "fragmented"
Fields
- id: Question identifier (e.g., "SQuID_0001")
- image: Satellite image path
- question: Question text with GSD notation
- answer: Ground truth answer
- type: One of 24 question types
- tier: Difficulty level (1=Basic, 2=Spatial, 3=Complex)
- gsd: Ground sampling distance in meters
- acceptable_range: [lower, upper] bounds for numeric questions (when applicable)
Evaluation
For numeric questions, check if predictions fall within the acceptable range:
import math
def evaluate(prediction, sample):
if 'acceptable_range' in sample:
# Numeric question - check if within human agreement range
lower, upper = sample['acceptable_range']
return lower <= float(prediction) <= upper
else:
# Non-numeric question - exact match required
return str(prediction).lower() == str(sample['answer']).lower()
The acceptable ranges represent the natural variation in human perception for spatial measurements.
Dataset Distribution
By Tier
- Tier 1 (Basic): 710 questions (35.5%)
- Tier 2 (Spatial): 616 questions (30.8%)
- Tier 3 (Complex): 674 questions (33.7%)
Top Question Types
- complex_multi_condition: 490 questions (24.5%)
- count: 178 questions (8.9%)
- binary_comparison: 172 questions (8.6%)
- size: 166 questions (8.3%)
- percentage: 157 questions (7.8%)
- proximity_percentage: 123 questions (6.2%)
- binary_proximity: 122 questions (6.1%)
- proximity_area: 107 questions (5.3%)
- connectivity: 104 questions (5.2%)
- fragmentation: 98 questions (4.9%)
By Source
- DeepGlobe (0.5m GSD): 612 questions, 174 images - Land use classification masks
- EarthVQA (0.3m GSD): 1241 questions, 364 images - Building detection and land cover masks
- Solar Panels (0.3m GSD): 97 questions, 35 images - Solar panel segmentation masks
- NAIP (1.0m GSD): 50 questions, 14 images - Human-annotated diverse scenes
Statistics Summary
- Zero-valued answers: 102 (5.1%)
- Questions with ranges: 1577 (78.8%)
- Average questions per image: 3.4
Notes
- Questions explicitly state minimum area thresholds (e.g., "ignore patches smaller than 0.125 hectares")
- Zero-valued answers indicate absence of features (intentionally included for robustness testing)
- The benchmark tests both presence and absence of spatial features to avoid positive-only bias
- Human agreement ranges allow for natural variation in spatial perception and counting
- All measurements use metric units based on the specified GSD (Ground Sampling Distance)
- Count ranges use proportional MADc (0.19) so larger counts have wider acceptable ranges
Source Datasets & Attribution
SQuID is constructed from publicly available remote-sensing datasets. We use only images from published validation or test splits and comply with the original dataset licenses.
DeepGlobe
Ilke Demir et al., DeepGlobe 2018: A Challenge to Parse the Earth through Satellite Images, CVPR Workshops 2018.
Source: https://deepglobe.org/EarthVQA
Junjue Wang et al., EarthVQA: Towards Queryable Earth via Relational Reasoning-based Remote Sensing Visual Question Answering, ICCV 2023.
Source: https://github.com/WangJunjue/EarthVQAPhotovoltaic (Solar Panels) Dataset
H. Jiang et al., Multi-resolution dataset for photovoltaic panel segmentation from satellite and aerial imagery, Earth System Science Data, 2021.
Source: https://essd.copernicus.org/articles/13/5389/2021/NAIP Imagery
U.S. Geological Survey, National Agriculture Imagery Program (NAIP).
Source: https://www.usgs.gov/core-science-systems/national-geospatial-program/national-agriculture-imagery-program
All derived annotations, questions, and acceptable answer ranges introduced in SQuID are released under CC BY 4.0.
Citation
If you use this dataset, please cite:
@misc{massih2026reasoningpixellevelprecisionqvlm,
title={Reasoning with Pixel-level Precision: QVLM Architecture and SQuID Dataset for Quantitative Geospatial Analytics},
author={Peter A. Massih and Eric Cosatto},
year={2026},
eprint={2601.13401},
archivePrefix={arXiv},
primaryClass={cs.CV},
url={https://arxiv.org/abs/2601.13401},
}
- Downloads last month
- 62